474
UWE R. MULLER et al.
ogies have evolved to print 3 to 6 thousand individual target spots per cm
2 ,
if each spot has a diameter of 80-100 Jlm and spots are separated by 50-80
Jlm from each other. Using microfabrication technologies pioneered by
the semiconductor industry, spot sizes of less than 10 Jlm are quite feasible, providing for over a quarter of a million different target sites in the
same area.
The intended use of the DNA chip determines its design with regard to
target DNA size and composition. Since for short oligonucleotides the specificity of the hybridization reaction at a given stringency is mainly a function of the length of the DNA, the conditions can be chosen such that a
single point mutation can be detected in the sequence to be analyzed
(herein referred to as probe). If multiple target sequences are used simultaneously, cross-hybridizations are inevitable and a different strategy has
to be employed. Based on this principle, chips can be designed with as
many as 65,536 (all possible 8-mer sequences) or even more target spots
on a single chip to unambiguously deduce the sequence of a specific DNA
probe. This process, called sequencing by hybridization (SbH) (Yershov et
a1.1996; Southern, 1996; Drmanac et a1.1998; Chee et a1.1996; Drmanac et
al.1989), generates very complex hybridization patterns and requires sophisticated computer software to interpret the image. In addition, the sequence to be analyzed must be fairly short and must be isolated and amplified from the rest of the genome through PCR (polymerase chain reaction) before it is applied to the chip for sequence analysis. Nevertheless,
SbH is a very powerful technique that lends itself to any application where
the exact nucleotide sequence has to be determined for an appropriate
diagnosis or therapy. It is less useful, however, for diagnoses of gross chromosomal changes or alterations in the level of gene expression.
Judged by the number of publications and commercial efforts, the emphasis in the past few years has clearly been on the development of arrays
for expression profiling. Both arrays of oligonucleotides as well as cDNAs
have been employed for that purpose (Zhu et al. 1998; Schena 1996; Noordewier and Warren 2001; Rew 2001), though they differ with respect to a
variety of funtional parameters. Expecting higher specificity and increased
power to discriminate between very similar genes and splice variants, the
trend is now shifting toward oligomer targets for expression applications
(Hughes et al. 2001).
Because of the demand for higher sensitivity and accuracy the development of arrays for genomic analyses started with chips containing BAC,
PAC, or PI clones, each containing an insert ofhuman genomic DNA (Pinkel et al. 1998; Solinas-Toldo et al. 1997). It is the size of these inserts,
typically ranging from 80 Kb to 150 Kb, and their spacing along the human
UWE R. MULLER et al.
ogies have evolved to print 3 to 6 thousand individual target spots per cm
2 ,
if each spot has a diameter of 80-100 Jlm and spots are separated by 50-80
Jlm from each other. Using microfabrication technologies pioneered by
the semiconductor industry, spot sizes of less than 10 Jlm are quite feasible, providing for over a quarter of a million different target sites in the
same area.
The intended use of the DNA chip determines its design with regard to
target DNA size and composition. Since for short oligonucleotides the specificity of the hybridization reaction at a given stringency is mainly a function of the length of the DNA, the conditions can be chosen such that a
single point mutation can be detected in the sequence to be analyzed
(herein referred to as probe). If multiple target sequences are used simultaneously, cross-hybridizations are inevitable and a different strategy has
to be employed. Based on this principle, chips can be designed with as
many as 65,536 (all possible 8-mer sequences) or even more target spots
on a single chip to unambiguously deduce the sequence of a specific DNA
probe. This process, called sequencing by hybridization (SbH) (Yershov et
a1.1996; Southern, 1996; Drmanac et a1.1998; Chee et a1.1996; Drmanac et
al.1989), generates very complex hybridization patterns and requires sophisticated computer software to interpret the image. In addition, the sequence to be analyzed must be fairly short and must be isolated and amplified from the rest of the genome through PCR (polymerase chain reaction) before it is applied to the chip for sequence analysis. Nevertheless,
SbH is a very powerful technique that lends itself to any application where
the exact nucleotide sequence has to be determined for an appropriate
diagnosis or therapy. It is less useful, however, for diagnoses of gross chromosomal changes or alterations in the level of gene expression.
Judged by the number of publications and commercial efforts, the emphasis in the past few years has clearly been on the development of arrays
for expression profiling. Both arrays of oligonucleotides as well as cDNAs
have been employed for that purpose (Zhu et al. 1998; Schena 1996; Noordewier and Warren 2001; Rew 2001), though they differ with respect to a
variety of funtional parameters. Expecting higher specificity and increased
power to discriminate between very similar genes and splice variants, the
trend is now shifting toward oligomer targets for expression applications
(Hughes et al. 2001).
Because of the demand for higher sensitivity and accuracy the development of arrays for genomic analyses started with chips containing BAC,
PAC, or PI clones, each containing an insert ofhuman genomic DNA (Pinkel et al. 1998; Solinas-Toldo et al. 1997). It is the size of these inserts,
typically ranging from 80 Kb to 150 Kb, and their spacing along the human